A compression encoding technique is used so as to efficiently transmit or accumulate moving picture data. In the case of a moving picture, MPEG1, 2, 4 or H.261 to H.264 systems are widely used.
In order to improve the encoding efficiency, an encoding-target picture is divided into a plurality of target blocks, and then predictive encoding/decoding processing is performed. Specifically, a picture for one frame is divided into target blocks of 16×16 pixels. In the intra-frame predictive encoding method, a predicted signal is generated using an already reproduced, adjacent picture signal (a picture signal uncompressed from compressed picture data) which resides within the frame in which the encoding-target block resides, and a difference signal is then encoded which is obtained by subtracting the predicted signal from the signal of the target block.
The inter-frame predictive encoding method may be additionally used, in which a prediction signal of the target block is generated by referring to other pictures adjacent on the time axis in a moving picture, and a difference between the target block and the prediction signal is encoded. In this case, a motion detection is carried out for the target block by using other encoded and restored frames as reference pictures, and a prediction signal associated with the smallest difference is decided, and then the difference value between the target block and the prediction signal is calculated. Then, discrete cosine transformation and quantization processing are performed on the differential signal obtained by the intra-frame prediction or inter-frame prediction. Encoded data is generated by entropy encoding of a motion vector or mode information for specifying a prediction signal and a coefficient of discrete cosine transformation quantized.
The picture data encoded in this way is restored and reproduced in order to acquire a prediction signal of a next target block, and is temporarily stored in a frame memory as a reference picture.
As described above, in the conventional encoding/decoding processing, prediction or conversion processing is performed while the bit length of each pixel of an input picture is fixed. In a normal picture, each pixel of brightness and color difference components is expressed with 8 bits of information. Accordingly, the encoding/decoding processing, such as prediction and conversion, is also performed on each 8-bit pixel. The reproduced 8-bit reproduction picture signal is stored in a frame memory.
On the other hand, a technique of reducing the size of a frame memory or a memory access band (memory band) for downsizing and reducing the power consumption is disclosed in Patent Literature 1 mentioned below. In this technique, a reproduction picture is additionally compressed before it is stored in the frame memory. Specifically, a reproduction picture reproduced with 8 bits is quantized to become one with bits less than 8 bits and is then stored in the frame memory. When generating a prediction signal, the compressed reproduction picture is expanded and used as a reference picture. Thus, since the encoding/decoding processing is performed with the bit length of an input picture being fixed, the compression encoding performance is prevented from being lowered while the size or memory band of a frame memory is reduced.